MUXA Design Studio
MUXA Design Studio
How we helped a narrative intelligence platform move users from an overwhelming stream of mentions into a structured, navigable map of the stories that actually matter.
PeakMetrics helps organizations track how narratives develop across the media landscape. The platform already had a Narratives feature, a way to group and label clusters of related mentions.
But as online conversations became more layered and fast-moving, they wanted to go deeper with their narrative offering. Analysts knew a narrative like "a major manufacturer's safety crisis" actually contained multiple distinct threads, a product defect, internal whistleblower testimony, a regulatory investigation, but it was difficult to find exactly that in the tool.
They were exporting data into spreadsheets, color-coding rows, and manually mapping sub-topics just to hold the structure in their heads.
We were brought in to redesign the Notifications experience and with the success of that work, we stayed on to work on the Narratives experience and extend it with a new concept: Layers. The goal was to give the existing narrative system even more depth, without breaking the workflows analysts already relied on.
PeakMetrics is an AI-driven narrative intelligence company that helps enterprises and governments see through the manipulated internet and protect against emerging adversarial threats. They uncover the online narratives shaping public perception, reputation, and business outcomes, revealing both risks and opportunities. The platform provides critical context into how narratives spread, why they gain traction, who is amplifying them, and whether they are authentic or engineered, so teams can act early and with confidence.
PeakMetrics already grouped mentions into Narratives, broad story buckets that helped users filter and track coverage. Each narrative is itself a cluster of individual mentions, the raw posts, articles, and broadcasts pulled from across news and social media that share the same story. Our redesign preserved the familiar flat narrative list, still the default for quick daily monitoring, and introduced an optional Layers mode, a hierarchical extension that lets users explore the structure across narratives, not just within them.
When Layers mode is enabled, the list view transforms into a spatial Bubble Map, where each narrative appears as a circle and its sub-narratives as smaller circles nested inside. Size encodes the volume of mentions; nesting depth conveys specificity. The result is an immediate visual sense that some stories contain worlds within them.
Users can toggle between flat and layered views, switching from speed to depth as their workflow demands. In the Bubble Map:
Together, Layers and the Bubble Map give users a dimensional understanding of media narratives, revealing not just what is being said but how each story branches and evolves over time.
To make the depth navigable, we added an abstraction slider at the base of the map. Move it left and sub-narratives collapse into their parent clusters; move it right and the full hierarchy expands. The user controls the resolution, the dashboard does not decide for them.
When a user identifies a narrative to investigate, the detail view gives it depth. This screen serves two cognitive modes: the analyst building a picture of the discourse landscape, and the communications professional looking for how to respond.
The AI-generated summary was the most iterated element. Early versions read like data reports. We worked to produce something closer to a briefing, leading with risk level, naming the specific channels driving the narrative, and surfacing the key messages being propagated.
The detail view brings several patterns together:
Analysts told us the filter panel was where they actually configured their workspace. We redesigned it as a real navigation surface, not a tucked-away modal: clear sections for Date Range, Channel, Domain, Narrative, Author, Language, each with counts so users could see what they were narrowing toward.
The Threat Score, a 0 to 10 AI-generated reputational risk signal, appears across the workspace overview, the narrative list, and the detail view. Getting it right requires solving a trust problem: teams use this score to decide which narratives deserve a response. A misread score has real consequences.
We made two key decisions. First, make the score legible, not just a number. Second, treat it as a starting point for investigation, not a verdict.
Extending an existing feature is harder than building a new one. Users have habits, mental models, and workflows already formed around the flat narrative view. Not everyone likes change, even if it is designed to make exploration more powerful.
The biggest design challenge was not the Layers feature itself, it was making it feel like a natural evolution of something familiar, not a replacement for it. Keeping the flat view as the default, and making Layers clearly optional and toggleable, was the decision that made adoption possible.
And on AI: transparency outperforms simplicity when the stakes are real. We could have shown just the number. We chose to show the number, the reasoning, and the timestamp, knowing some users would never open Contributing Factors and others would live in it. That investment in explainability was the right call for building trust.
Other patterns we developed during this work, like the synchronized map and list view, the filter panel redesigned as a navigation surface, and the onboarding pattern that respects users who do not activate optional features, are covered in our accompanying blog post.
Read, Designing for Depth: How We Redesigned PeakMetrics' Narrative Dashboard →